Quantifying custom genes abundance

I’m aiming to quantify the abundance of set of custom genes in a shotgun metagenomics sample.
I find both shortbred and humann2 suitable for this purpose.
While Shortbred seem to be developed for this exact purpose, HUMAnN supplies additional layers of information.
I’m trying to think which of the tools better suits while taking into account that my samples are pretty unique and I assume that a relatively high portion of the genes doesn’t exist in a databases such as uniref and hence a custom protein database should be supplied.
Thank you very much!